LZ: summary of the minutes of the tinet meeting


[Prev][Next][Index][Thread]

LZ: summary of the minutes of the tinet meeting



this is a summary of the meeting for the tinet. if you would like a
transcript of  the actuall minutes of the meeting please email me, my
address is at the bottom.










attendee's             :5
length                 : 2 1/2 hours
topic                  : protocols






discussion:
        1 )  type of protocol..
        2 )  safety
        3 )  implementation
        4 )  equipment
        5 )  one problem


        before reading on, i would like to thank the dillegent 5 who showed
up for the tinet meeting. although some of the people who seemed the most
excited about this didn't show up (clegg) :) and some of the origional
people in the tinet didn't show up, (zach) :) it was a good showing, and
some supprise people showed up and really took the show. without them, none
of this meeting would have really gone well. thanx to the guys with the
nicks: iliterate , duster picaman, and EliteSpy .. and yes i was there too
as NeTwOrKeR. well, we need lots of programmers now, so if you , after
reading this, want to help, please show up at the next meeting saturday at
10:00 est or edt which ever floats your boat (they both mean the same thing)
at irc.erols.com room #tinet. tanx




item #1     type of protocol :


        the type of protocol decided on was a packet switched token ring.
with this implementation, there is no trafic and garbled data, only 2
frequencies are needed, and people can join in and log off at any time. will
go into depth further in the implementation section. another bonus is that
the actuall protocol program doesnt have to be HUGE in order to work, it
could possibly be done in as little as 4k, we figure.


item #2    safety :


        safety. due to the fact that the packets will allow for private
conversation, privacy will be a biggie. to account for this, the source to
the protocol its self will not be relased, only the data requirements. the
programs themselves when made will have releaseable source code. besides,
most of you could figure it out pretty easily anyway... :)


item #3    implementation


        the actuall protocol will be set so only one calculator (the one
with the 'token' can talk at once. remember, listening doesnt cause trafic
only talking. when that particular calc speaks, the info is sent to all
calcs. there will be a header that tells the calcs what to do with that
particular info, ie if the number is the same display it, if it isnt dont.
the calculator has 2 trys at recieving the info. if no acknolegment is sent
back the info is discarded, and the token is passed on to the next number.
if there is no ack from the next number reciving the token, all calcs above
that particular number, (remember they can listen just not talk) will shift
one number lower to close the gap. then the token is passed. joining into
the circle is still a problem,  and there are two ideas towards getting past
that.  1)inside each 'packet' the number of people on the net will be
included, at which point a new user will assume the next number up. if two
join at the same time, a number will be randomly drawn and the higher will
get the debated number, and the lower will wait for the next one. 2) the
calc can watch for a full rotation in the token ring, which always goes in
order, then assume the next number, and the same thing can happen if 2 try
to get the same number. logging off will be easy because as soon as the
token gets to the missing number and no ack is recieved, the gap is filled
with a shiftdown by all numbers greater than that. if the calc with the
token has nothing to say at that moment it passes the token and waits for
the next one.




item #4       equipment:


all that will be needed is a standard 2 frequency rtlink (not the one that
is on ticalc.org) Duster (sorry i forgot your name) is going to work on a
new link along with my friend. in estimation, the cost should be about the
same for the new one as the old one. before we release anything, it will all
be tested, so as not to gyp everyone into buying parts to someting that
doesn't work. <hint guy who released plans to the rtlink>.




item #5        one problem:


        the 5 of us are still baby programmers, and wouldn't be able to
accomplish this. we need programmers. if you think you are pretty spiffy at
this and could help, we can really use you... i think that this summery is
fairly complete, and should answer most questions. if there are any, or you
are considering showing up this weekend, email me @
cbporter@erols.com 






           this has been a summary of the minutes of the tinet meeting




                                                        THANX
chris